题目：《Technical Note Performance evaluation of a small-animal PETCT system based on NEMA NU 4– 2008 standards》
1.论文摘要目的:MetisTM PET/CT是一种自主研发的、基于硅光电倍增管(SiPM)探测器的啮齿动物PET/CT系统。本研究的目的是使用国家电气制造商协会(NEMA) NU 4-2008标准协议评估系统的性能。方法:测量能量分辨率、空间分辨率、灵敏度、散射分数(SF)、噪声当量计数率(NECR)和图像质量(IQ)特征。利用三维有序子集期望极大化(3D-OSEM)和最大似然期望极大化(MLEM)重建图像进行了微Derenzo体模实验，评估了空间分辨率。此外，将CT显像剂Ioverol 350与fluorine-18(18F)-fluorodeoxyglucose (FDG)混合，注射到微型Derenzo体模中评估PET/CT成像。用18F-FDG对健康小鼠和大鼠进行了体内PET/CT成像试验。结果:该系统的平均能量分辨率为15.3%。视场中心的切向分辨率为0.82 mm半峰全宽大值(FWHM)，径向和轴向分辨率普遍低于2.0 mm FWHM。3D-OSEM的空间分辨率显著提高，特别是轴向半宽可提高约57%。当能量窗为200-750 keV和350-750 keV时，系统的绝对灵敏度分别为7.7%和6.8%。小鼠样体模的散射系数为8.2%，大鼠样体模的散射系数为12.1%。在69 MBq时，小鼠和大鼠样体模的NECR峰值分别为1343.72 kcps和640.32 kcps。IQ体模中的1mm可填充杆可以清晰地观察到。我们可以使用3D-OSEM(10个子集，5次迭代)清晰地识别出0.6 mm孔径的微Derenzo幻像。我们还对小鼠的PET和CT图像以及大鼠的脑成像进行了融结论:结果表明，该系统具有高分辨率、高灵敏度和高智商的特点，适用于基于啮齿动物成像的研究。
图一PET系统由32个检测块组成，这些检测块排列在四个相邻环中，环直径为129 mm（有效环直径为81 mm），轴向范围为122 mm。单个探测器块包含两个晶体模块，由24×24个晶体（每个晶体0.943×0.943×10 mm3）组成的阵列组成，晶体的中心距为1.028 mm。
The mean energy resolution of the whole PET/CT system is 15.3%, with the best and worst energy resolutions being 7.4% and 49.8% respectively.
The measured FWHM of 22Na point source using FBP in three directions are shown in Figure 3a. The FWHM resolutions are between 0.82 and 1.94 mm. The FWHM volumetric resolutions for the axial FOV and 1/4 axial FOV from center are 1.16 and 1.56 mm3 respectively. Figure 3b shows the FWHM of the point source after 3D-OSEM reconstruction with scatter and random corrections.
Figure 4 shows the PET images of the micro Derenzo phantom, and the CT image after injecting Ioverol 350 developer. The tangential resolution is the best when using 3D-OSEM (10 subsets, 5 iterations), between 0.6 and 0.5 mm. The radial resolution
is 0.7 mm, and the axial resolution is the worst, between 1.0 and 0.9 mm.
The absolute peak sensitivity of the system is 6.8% and maximum sensitivity is 0.062 cps/Bq with the energy window of 350–750 keV at the center of FOV, while the absolute peak sensitivity value is 7.7% and maximum sensitivity is 0.070 cps/Bq with the energy window of 200–750 keV. Figure 5 shows the sensitivity profiles measured along the system axially using the 22Na
The SFs for the mouse- and rat-like phantoms are 8.2% and 12.1% respectively. The total, true, random,and scatter coincidence count rates performance for two phantoms are shown in Figure 6. The NECR is 1343.72 kcps at 69 MBq for the mouse-like phantom, and 640.32 kcps for the rat-like phantom at 53 MBq, whereas the peak true count rate is 1666.94 kcps at 69 MBq and 1038.48 kcps at 113 MBq for mouse- and rat-like phantoms, respectively.
The PET images of the IQ phantom reconstructed using MLEM are shown in Figure 7a, with even the smallest (i.e., the 1-mm) rod visualized. The average, maximum, minimum activity concentration and the %STD in the uniform area are shown in Table 2. RCs and %STD for five rods of different size are shown in Figure 7b. The SOR is 0.20 (19.47%) for the water-filled chamber and 0.14 (17.75%) for the air-filled chamber.
Figure 8 shows coronal-section PET and CT images of a healthy mouse using PMOD software. Figure 9 shows a clear PET images of the healthy rat brain.
Different PET system designs and implementations will lead to differences in the sensitivity of systems. Compared with previous preclinical systems, MetisTM PET/CT system has a smaller scintillation crystal size, smaller detection ring diameter, and larger axial extent, which increases the solid angle coverage and therefore increases the system sensitivity.Owing to the increased sensitivity, the NECRs have also been significantly improved. The system peak NECR values are 1343.72 kcps for the mouse-like phantom and 640.32 kcps for the rat-like phantom, which are better than most commercial small animal PET/CT systems.